<bandali>i just installed Guix on the office terminals of our school’s cs club <bandali>next step hopefully: install it on all of the club’s servers :p <buenouanq>finally some abuse of power I can get behind *bandali does evil laughter <pkill9>is that guix system or guix on another distro? <bandali>the terminals run ubuntu and main servers debian <bandali>though i’ve heard that we might get a few new servers, and i’ll do my best to make sure at least one of them will run guix system ;) <ItsMarlin>can i start ssh server without configuring the scm file? <Ozymandias42>I think you can. just install the package yourself and either manually start it or tell shepherd to start the service. Not sure if you get the service automatically with the package and just not enabled. <Ozymandias42>it should have installed it. it might not have installed the service but you could just start sshd manually, I think <pkill9>sshd is in <openssh package>/sbin, which won't be in your PATH if you haven't installed a package with an sbin executable before <Marlin1113>does anybody know the right way to setup multiple keyboard layouts and do tweakings like changing ctrl with caps? <krafter>Can I set auto-login to only occur once? <krafter>The issue I have is that if X doesn't start properly then the PC will get stuck in a login loop that I can't get out of. <str1ngs> normally you can add single to kernel arguments in grub. to boot into single user mode. I don't know if this works with Guix but it might. <str1ngs>that might avoid your boot loop issue. <str1ngs>it boots into single user mode. meaning it does not start multi user services. like GDM networking etc <str1ngs>and I dunno if it works with Guix, but it's worth a try <krafter>Oh, you mean like a temporary workaround? <krafter>Ok, I was looking for a more permanent solution, but I guess its a useful tool to have in my toolbox. <str1ngs>I just tested single kernel arguement didnt seem to work. <str1ngs>though possibly rolling back to an known working profile is a better option. though that might not help your login issue ***jje_ is now known as jje
<bluekeys>Merlin113: I think there is a nocaps option that can be set in config.scm <bluekeys>I mean Marlin1113: There is a nocaps option that cna be set in config.scm <bluekeys>How do i get my wifi to work when I'm using stumpwm? Xfce has a network icon but stock stumpwm has nothing. <bavier`>bluekeys: assumming NetworkManager, you should be able to lauch 'nm-connect' or 'nm-tui' from a terminal <bluekeys>bavier: That's great. It works thank-you. <jonsger>nice. Guix system up and running now in Hetzner Cloud :) There were some caveats on the way... <kmicu>jonsger: ヽ(*＾▽＾)／ That sounds lovely. Thank you. <asweetmouse>Good morning. Can somebody help me with an Openvpn issue? <asweetmouse>Apparently it's not a good idea to run Openvpn as root, and I'm trying to see how to set it up where I don't have to run it every time. <asweetmouse>Just can't figure out how to do it, and I'm not smart with computers. <asweetmouse>I sent a message to the mailing list awhile ago, but so far no replies. <kmicu>Heh 30 May. Let’s give some time to Guix VPN experts 😺 *kmicu sees that openvpn-client-service-type extends shepherd-root-service-type. Guix can for sure do better. <Gamayun>Hm, I got a 'host not reachable' error for an email to guix-patches@gnu.org <asweetmouse>Let's just hope it's not too terribly difficult sorting this out. <Formbi>is it possible to suspend a GuixSD computer? <Ozymandias42>I don't know but I know what would be needed to answer the question with yes. You'd need a swap partition and your init-system had to support it because restarting services in the states they are when they are loaded back into memory differs from a clean boot. So I'd say if shepherd can do that and you have a swap partition it should work. <Ozymandias42>..wait I confused suspend with hibernate :D suspend is basically the same without the swap partition and easier. so I'd be surprised if that wasn't possible <Gamayun>Formbi: Suspending to memory should work out of the box. At least in desktop setups. Dunno about hibernate / suspend to disk. <Formbi>so how should I invoke it? «herd suspend»? <Gamayun>Formbi: I think elogind handles it... <Ozymandias42>does that handle acpi calls too? because I have a headless vm that can't be shutdown or restarted by the vm manager <janneke>hello civodul, i'm playing with emascy, remember that? ;-) <asweetmouse>So now I'm trying to add a supplimental group named "ip" to my /etc/config.scm to help with my Openvpn problem to start. <asweetmouse>guix system: error: supplementary group 'ip' of user 'ruki' is undeclared <raingloom>heyyy, i've been getting this error after a recent-ish upgrade: `error: failed to connect to `/var/guix/daemon-socket/socket': Connection refused` <Ozymandias42>I want to create a bash script that calls `exec /gnu/store/...shepherd --config /gnu/store/...shepherd.conf` and that is symlinked to by /sbin/init. The sylinking part works but I can't find anything about how to get the store path and what I tried from the gexp page in the docs or what works with extra-special-file won't work in the guile repl where I test it. <Ozymandias42>raingloom seems like you don't have access rights to that socket. try to correct that or remove it. it should get recreated automatically (hopefully) <janneke>civodul: i need to build a guix editor thingy with a browser -- obviously it needs to embed guile and i don't want to reinvent everything while reinventing emacs ;-) <janneke>emacsy looks pretty nice, i wonder why it looks like nothing really happened with it for years... <Tirifto>janneke: If you're reinventing emacs, aren't you already reinventing everything, practically? :o <janneke>Tirifto: i've tried several emacs branches that offer a webkit plugin, but they just don't work. embedding guile in a gtk3+ app is trivial and as a bonus i get a more modern lisp <civodul>janneke: what about str1ngs's Nomad? <civodul>maybe not ready for prime time, dunno <str1ngs>janneke: I am currently working on a extensible web browser using guile, C, C++ and QML . it's backend is qtwebengine <Ozymandias42>if you guys know scheme can someone maybe tell me how to query the path of a file in the store? I need that to generate a script with my system.scm <str1ngs>Ozymandias42: in bash the best way to do this is to do. guix build <package> <str1ngs>though the draw back to that is it could actually build the package :P <janneke>civodul, str1ngs -- thanks for mentioning that, I wasn't aware <Ozymandias42>I just need to get the paths to shepherd and shepherd.conf <str1ngs>janneke: is is very much WIP. but it could use more eyes. also the feature you are looking for is something we aim to add to nomad in the future. <Ozymandias42>it could easily be a guile script too but I am not experienced enough with that and I don't know how complex guile's equivalent to bash's exec is <str1ngs>janneke: though currently nomad does not access the DOM via scheme. it is feasible to wire java script and scheme together. I just have not done much in that direction yet. <janneke>str1ngs: i've been using next as my main browser... <janneke>str1ngs: i'd have to get over my hatred of qt, not sure how long that would take ;-) <str1ngs>most of Nomad is scheme, C, and QML . actually not that hard to hack on *janneke suffered another qt5 rebuild a couple of days ago <janneke>good thing you don't depend on qt, i'll maybe have a look then! <str1ngs>I've been rebuilding qtwebengine so I feel ya :P *kmicu has zero Qt in the system to avoid those painful compilation times. <str1ngs>janneke: I provide a substitute server for qtwebengine. so rebuilds are minimal <str1ngs>janneke: so nomad still depends on QT. qtbase and qtwebengine are the larger inputs <str1ngs>the alternative would be GTK webkit. and it has some technical drawbacks. <str1ngs>I actually started with GTK Webkit. but is DOM introspection is not the best for guile. guile likes to manage threads. and gtk likes to manage it's own threads. and literally you need to use dbus to interact with the DOM. with QT the GUI can be handles with simple QML. and the DOM can be accessed with QWebChannel . basically allowing you to wires javascript to C methods easier. sorry for the long technical explanation :P <janneke>str1ngs: no thanks! oh, that's really awful and good to know... <janneke>str1ngs: also, i'm glad that you started out with gtk and saddened that guile integration/support is so bad, apparently. <str1ngs>guils is amazing. the issue is more GTK IMHO <str1ngs>janneke: well more the combination of GTK and WebKit <str1ngs>janneke: and I think with sxml and qwebchannels . I think the application you would like to create is much more feasible. and I think is the direction I'll move most of Nomad's UI eventually. this should offload most things to scheme. <janneke>str1ngs: it would be nice if code could be shared, that's what i like about emacsy; it could be such a nice reusable library <str1ngs>janneke: nomad also has a reusable scheme library. but it's still dependent on C bits. which is hard to get around. even Emacsy would also do that. <civodul>dutchie: could it be that GUIX_LOCPATH is not set in the environment of the bash that's complaining? <civodul>can you try "cat /proc/PID-OF-THAT-BASH/environ | xargs -0 echo"? <dutchie>that bash isn't interactive, it just appears if i run `guix pull` (i suspect it's guix build forking a bash) <dutchie>oh hang on, maybe I didn't install locales for the root user, could that be it? <dutchie>i hadn't, but installing them doesn't seem to have fixed it <civodul>dutchie: in that case, could you post the complete command and its output? <dutchie>(it's still building the derivation) <civodul>dutchie: i think it's from the libexec/guix/substitute script of guix <civodul>so make sure GUIX_LOCPATH is set for the daemon itself <civodul>i believe the .service file does that, but you have to "guix install glibc-locales" as root <dutchie>might try a full reboot in a second i guess <civodul>but make sure that GUIX_LOCPATH appears in /proc for guix-daemon <civodul>and does GUIX_LOCPATH point to the right thing? <civodul>i mean the GUIX_LOCPATH as it appears in /proc <dutchie>yeah, GUIX_LOCPATH=/var/guix/profiles/per-user/root/guix-profile/lib/locale <civodul>could you "strace -f -o log -p PID-OF-GUIX-DAEMON" and run "guix build whatever"? <civodul>to see exactly what happens when that message is emitted <dutchie>ok, it's too big to put on a pastebin, just trying to find the relevant part <civodul>yes, typically a few lines before the warning message <dutchie>oh, it's trying to open en_US.utf8 instead of .UTF-8 <dutchie>there's a lot of places in the guix source where it's .utf8 but I've only ever seen .UTF-8 and that exists in GUIX_LOCPATH/2.28 <jonsger>why is guix (the command) itself still on 0.16.0-13 when I install from the 1.0.1 image <jonsger>ah, on Guix system you need to do the path dance as well :( <civodul>dutchie: dpaste.com is blocking Tor users apparently, can you paste it elsewhere, like paste.debian.net? <civodul>dutchie: could you paste the ~50 lines that end with: write(2, "blah en_US.utf8 blah")? <dutchie>there aren't any: grep 'write(2, ".*en_US.utf8' gives no results <dutchie>if i just grep for en_US.utf8 it only gives calls to openat <civodul>well, maybe add "-s 300" to the strace command line <dutchie>(i'm going to have to head out soon, i'll collate this into a reply on the ml) ***ym555_ is now known as ym555
<asweetmouse>Welp, I'm at a loss as to what to do for now. I really don't understand scheme or how to modify system services to get Openvpn running. <Ozymandias42>same for me. I want to query paths from the store in my system.scm but it's not documented how to do it. <Ozymandias42>I probably need to search through guix's source to find how ***Swedneck2 is now known as Swednec4
***Swednec4 is now known as Swednec6
***Swednec6 is now known as Swednec8
<asweetmouse>@strings: yes. I'm trying to use a .opvn file without running Openvpn with root access like it wants to. ***Swednec8 is now known as Swedne10
<str1ngs>asweetmouse: you can use openvpn-client-service with openvpn-client-configuration ***Swedne10 is now known as Swedne12
<str1ngs>though I don't know if handles opvn files ***Swedne12 is now known as Swedne14
<asweetmouse>Well, all I'd like to be able to do is run the opvn file as non-root user. Since I'm having to run it as root already. Do I add it to my config.scm file? How to? <str1ngs>actually this may not work if you want to run unprivileged <str1ngs>replace guix-publish with openvpn-client <str1ngs>and then use openvpn-client-configuration fields <str1ngs>maybe you can translate your opvn file to configuration fields. alternatively you might want to make a shepherd user service. this would allow you to use opvn file and run unprivileged. <str1ngs>also you can maybe use networkmanager and network-manager-openvpn <rekado_>civodul: I’ll really need to change mumi to fetch emails from the debbugs HTTP frontend instead of the SOAP service. It’s really ugly to do this, but the SOAP interface is really just not up to the task :-/ <rekado_>so mumi will become a web interface to a web interface. Ugh. <asweetmouse>I already have network-manager and network-manager-openvpn installed. Running the gnome3 proxy settings tool doesn't work with the opvn file. I think it has something to do with the formatting of the file. <mbakke>jonsger: Guix will downgrade itself if you never run 'guix pull' before reconfigure. <str1ngs>asweetmouse: maybe someone else knows more about openvpn can help. my knowledge is kinda limited on that <asweetmouse>well, I guess I'll just keep waiting on the mailing list for now, or until somebody else can help here. ***Swedne14 is now known as Swedne16
<bluekeys>Is there a preferred way to use xmodmap to map keys? Can some config be passed to config.scm? ***Swedne16 is now known as Swedne18
<civodul>rekado_: in many cases mumi does display patches though, so perhaps there's something special about this one? ***Swedne18 is now known as Swedne20
<civodul>it uses XKB specs, which might be less expressive than xmodmap <bandali>civodul, is it possible to use services like dovecot through guix on a foreign distro? <rekado_>civodul: no, it’s really a problem with the SOAP service. It truncates certain emails so that some parts are lost. <rekado_>we’ve seen this on some other messages, unfortunately, and there’s an old bug report about this with no solution <rekado_>it’s not like it couldn’t be fixed, it’s just that the maintainer hasn’t been convinced that it’s worth fixing. <rekado_>the official solution is to … use the web interface to get at the raw emails. <rekado_>bandali: for services you need a Guix system, but you don’t need to have it installed as the only distro. You could, for example, run it in a container / a VM. <bandali>rekado_, oh cool! do i normally need root to run containers/vms on foreign distros? <civodul>we should consider running our own instance <bandali>my use-case is, i’d like to be able to run a local dovecot on a non-privileged port as a non-root user <rekado_>civodul: maybe. It might already be better with the upstream variant. The GNU fork is kinda strange. <rekado_>(but it’s all Perl… so I don’t really want to host it) <civodul>that'd be a first step towards a rewrite in Guile, of course :-) <civodul>at some point you won't resist the temptation *rekado_ feels a tingling sensation <raingloom>uuh i tried getting rid of an error and now the guix-daemon systemd unit doesn't start, the error is: Unbound variable: %storedir <raingloom>i tried linking current-guix to some previous versions buuut it didn't change anything <raingloom>is there a way to sort of reinstall it (on top of Arch) without losing my settings? ***Swedne20 is now known as Swedneck
<cbaines>raingloom, that sounds like the guix-daemon.service file just needs tweaking <raingloom>i looked at the ones in the guixes in the store and they are all the same <cbaines>So, you're running Arch right? Is /etc/systemd/system/guix-daemon.service a regular file, or a symlink? <cbaines>civodul, if you're around, I'm looking again at the linting stuff, and how it uses Gettext. You suggested making all the messages English, and translating later, but I'm not sure that'll work with the messages with data in them, e.g. (format #f (G_ "invalid description: ~s") description) <b0f0>to get tle ldd command, do I need libc or what do I need ? <cbaines>b0f0, gcc-toolchain is probably the thing to go for <b0f0>I still need init.el to configure packages for emacs. 'guix install some-emacs-package' just downloads them, to enable them in emacs, I need to enable the package in init.el. Do I understand this or not ? <cbaines>b0f0, I probably understand Emacs less than you, but for me, installing things is sufficient <cbaines>You might need to restart Emacs for it to pick them up <cbaines>There's also M-x guix-emacs-autoload-packages that can make them available without restarting <b0f0>how do I check if emacs has dbus support ? I should use ldd $(which emacs) and that should give me somthing but it doesnt ? <cbaines>b0f0, you'll have to dig through the wrapper scripts to find the actual executable <cbaines>It'll look something like /gnu/store/b38...m0q-emacs-26.2/bin/.emacs-26.2-real <cbaines>Or just do guix build emacs, and run ldd against the relevant file in the package <b0f0>ok so emacs has libdbus when run ldd on .emacs-26.2-real . This means my error when I build emacs-weachat package is really weird. Because when I run 'guix build -f emacs-weechat.scm' I get an error that says 'Emacs not compiled with dbus support'. <cbaines>b0f0, there might be multiple variants of the Emacs package <cbaines>b0f0, which variant are you using as an input to your emacs-weechat package? <cbaines>b0f0, this'll be in native-inputs, inputs or propagated-inputs in the package definition <cbaines>something like: ("emacs" ,emacs-minimal) <cbaines>it's the bit after the , that's key here <b0f0>"emacs-s" ,emacs-s) and ("emacs-tracking" ,emacs-tracking) are in propagated-inputs <cbaines>The relevant input will have "emacs" as the first bit of the lsit <cbaines>Hmm, the examples that I'm looking at that specify Emacs as a input use the gnu-build-system <cbaines>I guess you're using the emacs-build-system <b0f0>I am using emacs-build-system <b0f0>am I doing it al lwrong ? <cbaines>No, you're pretty much there, have a look at the emacs-auctex package <cbaines>The default "emacs" package that the emacs-build-system works with is emacs-minimal, that probably doesn't link against dbus <cbaines>But it's possible to specify the emacs package as a argument (#:emacs) to the build system <cbaines>(which the emacs-auctex package looks to do) <cbaines>The comment above (arguments seems similar <b0f0>you know how to do this cool, thank you very much for helping me <rekado_>raingloom: make sure you’re using the latest version. For a few hours there was a bug that caused the daemon to fail. <rekado_>(this reminds me: we should update the “guix” package) <cbaines>I think rekado commented that there was a momentary bug in Guix that you may have hit <cbaines>If the guix-daemon is running at the moment, running guix pull as root may resolve the issue <raingloom>and the install script refuses to install a new version since one is already present <cbaines>Right, so hopefully you can use the daemon from an older generation to at least run guix pull <cbaines>What files called current-guix-... do you have in this directory /var/guix/profiles/per-user/root <cbaines>Ok, great, so you have some older generations <cbaines>You can either switch the current-guix symlink to point at an older generation, or just run the guix-daemon as root temporarily in a shell, e.g. /var/guix/profiles/per-user/root/current-guix-1-link/bin/guix-daemon --build-users-group=guixbuild <raingloom>hmm. switching the symlink and restarting the unit results in the same error but so far there are no errors when i run it manually <cbaines>raingloom, if you get the guix-daemon running, I'd run guix pull as root <raingloom>yup, did that, stopped the manual one, started it up with systemd, and now it works